[Ffmpeg-devel-irc] ffmpeg.log.20131003
burek
burek021 at gmail.com
Fri Oct 4 02:05:01 CEST 2013
[00:13] <elBradford> Compiling with librtmp, I'm getting errors like "undefined reference to 'RTMP_Pause'" and for other rtmp functions. librtmp is building fine
[00:13] <elBradford> A response to a similar error message online was "This seems like librtmp was found at configure time and possibly compile time), but not at link time. Your installation of librtmp may be borked, e.g. by two conflicting versions." <--- but I'm not sure what that means
[00:17] <llogan> elBradford: what does librmtp offer that isn't support by native rtmp protocol support?
[01:21] <elBradford> llogan: -jtv support, which is required for my project
[02:37] <llogan> elBradford: sorry, but i don't know what -jtv is.
[02:44] <luosheng_> hi there, how can I get ffmpeg code at r20544?
[03:40] <elBradford> llogan: that's OK, it's just important that I have librtmp. Thanks for the response though.
[03:48] <relaxed> DaCoder: ffmpeg has the tee format, which I told you about earlier
[03:49] <relaxed> libav does not
[03:50] <relaxed> (sorry, wrong channel)
[03:55] <llogan> relaxed: i was going to tell him to add -pix_fmt yuv420p for his black screen issue but he vanished earlier
[03:56] <relaxed> add to the topic, for web use -pix_fmt yuv420p
[03:56] <relaxed> There's needs to be a very simple faq
[03:57] <relaxed> which points to more detailed info
[03:58] <relaxed> For the most part developers write the docs and they're horrible at it
[03:58] <llogan> the wiki is helping, IMO
[03:58] <relaxed> look at the split of ffmpeg's man pages
[03:59] <relaxed> there's ffmpeg-all not, but tstill
[03:59] <relaxed> still*
[03:59] <relaxed> now*
[03:59] <llogan> if you have a list of simple faq questions and answers I can help you implement something
[03:59] Action: relaxed has had too many pints
[04:00] Action: llogan has not had enough
[04:00] <llogan> i had a terrible IPA
[04:00] <relaxed> Regardless, the two of us could write much better docs
[04:01] <llogan> writing docs is probably harder than most people assume
[04:02] <llogan> it's definitely more time consuming than you'd expect
[04:02] <relaxed> It's because they're based on the old shitty docs
[04:02] <llogan> but if you're serious about contributing I will definitely help
[04:02] <relaxed> trust me, I've been reading them for years
[04:03] <relaxed> I think their whole approach is wrong
[04:04] <llogan> it's a big, complicated, very active project with many contributors with users using various versions with a bunch of shitty, outdated misinformation on blogs and forums. in short: it's not easy and there is always room for improvement.
[04:04] <llogan> and we have a problem with verbosity in some places and lack of info in others
[04:05] <llogan> what should the approach be?
[04:05] <relaxed> Splitting up the man pages wasn't a great idea. More examples in the man pages are needed.
[04:06] <llogan> are you willing to provide patches?
[04:07] <relaxed> I'm willing to bitch and maon
[04:07] <relaxed> which, I agree isn't that helpful
[04:07] <llogan> if you provide some examples I guess I can make some patches
[04:08] <llogan> once I review Timothy's patch that I've been ignoring since I got back to computerdom.
[04:08] <relaxed> I'm still torn between the two projects
[04:09] <llogan> that must suck. so easy for me only dealing with one.
[04:09] <llogan> seriously, sititng on the fence but be a pain in the taint
[04:13] <llogan> ping me if you want some help making some improvements
[04:13] Action: llogan has to go buy dog food
[04:14] <relaxed> Will do.
[04:39] <tempus_fol> Hello, I've tried to submit this question few days ago, I hope someone can help. It's on a Fedora 19. I'm trying to compile ffmpeg with libfdk_aac. I've added --enable-libfdk_aac to my .configure, and now it doesn't seem to build anymore (ERROR: libfdk_aac not found); latest version from Git for ffmpeg. I've tried rebuilding libfdk_aac from src.rpm _and_ from git (as per https://trac.ffmpeg.org/wiki/CentosCompilationGuide )
[05:23] <tempus_fol> Eventually, I've figured it out. For future refence: --extra-ldflags="-L/path/to/fdk-aac/lib" is needed (both libfdk and ffmpeg are compiled with --disable-shared (and ffmpeg with --enable-static))
[06:02] <ripthejacker> I have and video file in mp4 container, video codec h264 and audio codec aac. To what format do I need to convert it so that it would be lossless?
[06:03] <ripthejacker> I mean can I convert to any open source formats or should I stick to aac?
[06:03] <sacarasc> Why do you want to convert it, ripthejacker?
[06:03] <ripthejacker> It's video and I want just the audio
[06:04] <sacarasc> ffmpeg -i input.mp4 -vn -c:a copy output.mp4
[06:04] <sacarasc> I'd go with that, all it really does copy it without the video.
[06:05] <ripthejacker> sacarasc: so converting to any other codec will degrade the quality?
[06:05] <sacarasc> Yes.
[06:05] <sacarasc> Unless you went to something like flac, but that would increase the size a lot and not gain anything.
[06:06] <ripthejacker> sacarasc: what about vorbis?
[06:06] <sacarasc> That is a lossy codec, you would lose something.
[06:06] <ripthejacker> sacarasc: even with aac it will be still be lossy, right?
[06:07] <ripthejacker> *will still be
[06:07] <sacarasc> The lossiness has already happened, if you copy it (as in no extra encoding) it won't lose anything more.
[06:08] <ripthejacker> sacarasc: oh so to decode to another format it will lose some other data?
[06:08] <sacarasc> Yes.
[06:08] <ripthejacker> sacarasc: gotit
[06:08] <ripthejacker> sacarasc: thanks friend :)
[06:08] <sacarasc> Unless it is a lossless codec, like flac, but as I said, it will make the size a lot bigger.
[06:09] <ripthejacker> yeah I understand
[09:36] <BoR0> hi, I have a file that is 2GB. I want to stream it over RTMP with ffserver. I managed to stream with ffserver through HTTP port 8090, but how can I make ffserver listen RTMP?
[09:36] <BoR0> or does ffserver require separate RTMP server?
[09:51] <JEEB> BoR0, yes, you are supposed to use ffmpeg in conjunction with a separate rtmp(e) server
[09:52] <JEEB> that then handles the distribution
[09:52] <BoR0> is it the same case with RTSP?
[09:52] <JEEB> no idea to be honest
[09:52] <BoR0> can you suggest some good and free RTMP servers/
[09:53] <JEEB> there are a few that pop up every now and then but I have no idea how good they are
[13:25] <dum> hi all, is there any way to know what frame type the incoming h264 nal unit contains before calling avcodec_decode_video2?
[13:25] <dum> In some situations, I need to restart my decoder and I only want to start processing after receiving the Iframe
[13:25] <JEEB> usually containers tell you that info (random access point flag of one sort or another)
[13:25] <JEEB> but otherwise you'll just have to parse the NAL unit yourself
[13:41] <dum> JEEB: thank you. what is the best way of handling this situation? I am sure that parsing the NAL unit is not the common way right?
[13:42] <JEEB> the common way is to have the stream in a container and get the info from there
[13:42] <JEEB> since random access points generally are noted on the container level
[13:42] <dum> I transfer the naked NAL units always
[13:43] <dum> so you suggest me to parse it myself to handle this
[13:43] <JEEB> I don't really see another way in that case
[13:44] <JEEB> since the h264.c demuxer in libavformat doesn't really give much info and that stuff is then only parsed in the decoder
[13:44] <dum> can I find somewhere how to parse and what byte ro read for this info?
[13:44] <JEEB> you could check libavformat/h264.c and friends of course
[13:45] <dum> allright thanks a lot
[13:46] <JEEB> parsing of NAL units can be found in the H.264 specification
[13:46] <JEEB> http://www.itu.int/rec/T-REC-H.264-201304-I/
[14:46] <caraculo> hi
[14:50] <caraculo> the example for video encoding at "https://github.com/FFmpeg/FFmpeg/blob/master/doc/examples/decoding_encoding.c" produces a video, but I can not reproduce it in any player.
[14:50] <caraculo> I got "nvalid data found when processing input"
[14:51] <caraculo> I just compiled it, an ran it, exactly as they have it.
[14:52] <caraculo> yhe thing is that I have some code for FFv1 encoding, that seems to work fine, encodes without errors, but when trying to play. I get the kind same error
[15:06] <xlinkz0> i'm not understanding this : For constant frame rate videos a value of 1/2*frame_rate should address the worst case mismatch between the specified time and the time set by force_key_fram
[15:07] <xlinkz0> what do i need to do to make it more accurate?
[15:07] <xlinkz0> i've had tests where i've requested a timestamp at 0.5 seconds and it delayed it by 100ms, there were a lot of frames in that 100ms
[15:20] <caraculo> so It seems I have to put the FFv1 packets into a Container
[15:20] <caraculo> and I can not save them straight away into a file?
[15:23] <Mavrik> caraculo, you have to put pretty much all video and audio formats into some kind of container
[15:27] <caraculo> ok I ignored this,
[15:27] <caraculo> just that I saw the basic video_encoding example and I thought that replacing CODEC_ID_MPEG1VIDEO with CODEC_ID_FFV1 would do :p
[15:28] <caraculo> would you Mavrik please pass me an example on how to use ocntainers?
[15:29] <Mavrik> there's a muxing example.
[15:29] <Mavrik> filtering example does that as well IIRC
[15:30] <caraculo> Is it possible to just save packets without any container and then tell mplayer or vlc the decoder to use?
[15:30] <caraculo> ok thanks.
[15:31] <Mavrik> no, because you're missing data then
[15:31] <caraculo> ok thanks
[17:32] <cougarten> hi, i want to glitch arround some video. Can you point me in a general direction on how to gain control over the codecs work? I don't know any fitting channel but this here. Is ogg-video open-souce? I'd love to mess with the keyframes on an encoding level rather than destroing them via search+replace afterwards or something.
[17:33] <cougarten> ah, #vorbis. I'll ask there aswell
[17:39] <caraculo> If I pass an ".avi" extension file name to " avformat_alloc_output_context2" and it still can not deduce output format from file extension
[17:41] <durandal_1707> called av_register_all ?
[17:43] <caraculo> lol
[17:45] <caraculo> yes I do
[17:45] <caraculo> the thing is that this: " avformat_alloc_output_context2(&_oc, NULL, "mpeg", fname.c_str());"
[17:45] <caraculo> seems to fail to alloc
[17:47] <caraculo> will set _oc to NULL....
[18:34] <caraculo> I will bother you with my questions tomowwor... see you.
[23:11] <MonkeyFisto> Has anyone here used overclocking to improve ffmpeg speed?
[23:12] <MonkeyFisto> I found the forum to be rather bare so I came here to ask
[23:13] <MonkeyFisto> *cough*
[23:15] <radiaku> @monkeyfisto why you want to improve ffmpeg speed? I think its already fast :D
[23:15] <MonkeyFisto> Well I am encoding bluray rips
[23:15] <mittens> i dont think overclocking is worth much
[23:15] <MonkeyFisto> at stock settings that takes over 24 hours
[23:15] <mittens> unless you spend a lot on super cooling
[23:16] <MonkeyFisto> I ask because I got no gains at all
[23:16] <MonkeyFisto> I want to see if other got the same result
[23:16] <MonkeyFisto> *others
[23:16] <mittens> how much did you overclock by?
[23:16] <radiaku> what is your specs?
[23:16] <MonkeyFisto> 20%
[23:16] <MonkeyFisto> 4770k OC to 4.4GHz
[23:17] <MonkeyFisto> Stock averages about 3.7-3.8GHz
[23:17] <MonkeyFisto> after turbo
[23:17] <mittens> how many threads do you use for ffmpeg?
[23:17] <MonkeyFisto> Default
[23:17] <MonkeyFisto> whatever that is
[23:17] <mittens> 1 i think
[23:17] <mittens> how many cpu cores you got?
[23:18] <MonkeyFisto> well all my cores were maxed
[23:18] <MonkeyFisto> 8 logical
[23:18] <MonkeyFisto> 4 physical
[23:18] <mittens> try with 8 threads and see how that goes
[23:19] <MonkeyFisto> should it really matter though? Overclocking should improve encode speed despite the number of threads
[23:20] <mittens> all you probably did was make your cpu hotter
[23:20] <MonkeyFisto> So it would seem which is why I am asking here to see if others had a different experience
[23:21] <MonkeyFisto> I don't want to put forth the effort to do hours of testing if I am not going to get ANY improvement
[23:21] <MonkeyFisto> Increasing the CPU clock should increase the speed of all threads
[23:21] <MonkeyFisto> So if I run with 1 on OC and run with 1 without OC
[23:21] <MonkeyFisto> I should still see a performance increase
[23:22] <MonkeyFisto> Running with 8 will net me some gain
[23:22] <MonkeyFisto> but that is over 1 thread at the same speed
[23:22] <MonkeyFisto> If I get no increase using 1 thread and overclocking
[23:23] <MonkeyFisto> then it is unlikely that I will get an increase using 8 and OC rather than 8 and stock
[23:23] <MonkeyFisto> Though frankly I am surprised that ffmpeg doesn't auto-detect thread-based optimization :/
[23:25] <radiaku> Hey everyone, when I joining video with image using itsoffset, why ffmpeg is dropping frame? ( my video is 25fps )
[23:26] <fazias> ffmpeg as about everything else sane always uses good amount of threads
[23:27] <fazias> as it definately regocnizes your cpu on the fly and also decides on the fly which cpu optimizations can be used
[23:27] <radiaku> here my command just in case : ffmpeg -itsoffset 5 -i raisa.mp4 -r 25 -loop 1 -i bg.png -filter_complex "[1:v] fade=out:125:25:alpha=1 [intro]; [0:v][intro] overlay [v]" -map "[v]" -map 0:a -acodec copy out.mp4
[23:27] <fazias> setting a fixed thread count would be wrong atleast when using x264
[23:28] <fazias> or well, you would just get a bit lesser performance as things dont really work like, logical cpu count == threads
[23:30] <fazias> of course overclocking should increase performance, that is if you don't hit any other limits. Such like memory bandwidth(mmm unlikely), harddisk read/write (well you *can* hit a wall here, speed shouln't be problem anymore then tho)
[23:31] <MonkeyFisto> yeah neither of those things were an issue
[23:41] <teratorn> hmm, can anyone tell me how you would go about using a testsrc avfilter in C code? I basically need some video content for a test-case and found the testsrc filter... but it's not obvious how to use it?
[00:00] --- Fri Oct 4 2013
More information about the Ffmpeg-devel-irc
mailing list